Task scenario: 1. Analyze the data, clean the data, transform if needed. Provide a few key visuals. 2. Prepare a graph that shows the distribution of total number of minutes for international calls geographically in breakdown for evening and day calls. (It should be just one graph) 3. Make your assumptions on the relevant features for churn prediction. What extra features would you like to have, as you believe, they are crucial to explain churn. 4. Generate new features that are relevant to explain churn. For example: number of call per day for a particular person as compared to number of calls for a comparable sample (i.e. for the same state). 5. Select a model, justify your choice, build a model to predict churn. (Auto ML shall not be used) 6. Explain your model results, explain which features are important, which are not. Were your assumptions correct? If not, why do you think that is the case? 7. How would you test the model results? What is the best practice? 8. Present the results (showcase the model, explain each step, if needed support your presentation with power point slides) and insights that you managed to extract out of the data. 9. What if you had an access to Azure synapse workspace, how would you approach the task? How would you automate the model run? General Remark: You will not be judged solely by the final model quality, but rather by the quality of each task fulfillment. In real life it happens that model just don't work as expected, so as specialist be ready to explain WHY and what is needed to improve the results!